The first-administration record · Published · Experience only, never exam content
The First NextGen Bar Exam: What Actually Happened
2,624 people sat for the first administration of the NextGen UBE in late July 2026. In one state, 645 of them never got to start. Here is the complete record, every claim labeled by how well it's supported.
Two rules govern everything But For publishes about exam administrations. First: experience, never content. Examinees are bound to confidentiality about what the exam asked, and we hold ourselves to the same line, voluntarily and absolutely. We do not solicit, record, or repeat exam content. Second: every claim gets labeled. Official statements are marked as official. Press reports are marked with how many independent outlets carry them. Single reports are marked as single reports. Where we're inferring, we say so. The first week of a new licensing exam is a rumor-rich environment; someone should be the boring source that's right.
The administration by the numbers
Per the National Conference of Bar Examiners' official statement: approximately 2,624 applicants were scheduled across 10 jurisdictions and 16 testing locations for the inaugural NextGen UBE, a 9-hour exam over a day and a half, delivered on examinee laptops through NCBE's secure platform at jurisdiction-managed sites. OFFICIAL Scheduled headcounts in the three largest mainland jurisdictions, per NCBE's chief communications officer: roughly 800 in Maryland, 700 in Washington, 650 in Missouri. OFFICIAL, via ABA Journal Day 2 ran to completion with no new incidents reported. OFFICIAL
Seven of the ten jurisdictions (Connecticut, Idaho, Oregon, and the Pacific jurisdictions among them) appear to have administered without reported incident. OUR INFERENCE: no official or press report of disruption exists for them; absence of report is not proof of a flawless day.
Three jurisdictions had a harder Tuesday.
Washington: the exam that never began
The short version: Washington canceled its entire July bar exam without a single question answered by its standard-time examinees.
The sequence, assembled from the Washington State Bar Association's announcement, NCBE's statement, and consistent press reporting:
Tuesday morning, examinees at the Yakima Convention & Event Center, 645 standard-time takers by counts carried across three outlets PRESS ×3, could not begin Day 1. What followed was roughly five to six hours of attempted fixes, restarts, and waiting PRESS + examinee-side accounts relayed by secondary sources. NCBE describes the cause as an on-site hardware and configuration issue; the venue itself, in a statement to local press, called it a "catastrophic equipment failure." OFFICIAL, both By late afternoon, examinees were dismissed. Tuesday evening, the WSBA concluded it lacked sufficient confidence the problem was resolved and canceled Wednesday, and with it the administration. Its announcement said plainly: "We sincerely apologize to the candidates who worked so hard to prepare." OFFICIAL
One group was spared: examinees testing with approved accommodations (roughly 30 to 40 people) proceeded on their separate setup and completed the exam. PRESS ×3 + WSBA communications office
Affected applicants now choose among three options: a make-up administration tentatively set for September 1–2, a transfer to the February 2027 administration, or withdrawal with a refund of jurisdiction fees, with NCBE separately addressing its technology fee. OFFICIAL Press reporting indicates a decision deadline of August 14; affected applicants should confirm dates and terms against WSBA's direct communications rather than any news account, including this one. PRESS, single outlet on the deadline
The institutional fallout began within a day: at least one Washington law school has publicly urged diploma privilege for the affected class and asked the state supreme court for a transparent review, and press accounts describe early talk of litigation among examinees. PRESS, single outlets; developing, and we'll follow it
Maryland and Missouri: disrupted, completed
The same Tuesday produced the opposite outcome in two other jurisdictions. In Maryland, one testing site was impacted for under an hour; Day 1 completed on schedule. OFFICIAL In Missouri, site-specific network bandwidth problems delayed the start for many standard-time examinees; testing resumed the same day, and Wednesday's schedule was modified to compensate. OFFICIAL + local press
That contrast is the honest frame for the week: identical exam, identical platform, three very different Tuesdays, separated by the resilience of local infrastructure.
Platform versus plumbing
NCBE's central claim deserves precise treatment, because the next two years of bar-exam discourse will be built on it. The claim: the week's failures were site-local (venue networks, on-site equipment, local configuration) and "did not involve the NextGen testing platform itself." Where local infrastructure held, NCBE says, the platform performed as designed. OFFICIAL
Read fairly, the evidence is consistent with that claim: seven jurisdictions ran clean, two recovered same-day, and the catastrophic failure traces, by the venue's own account, to the venue. Read from a chair in Yakima, the distinction is academic: the examinee experience does not itemize whose equipment failed. Both readings are true, and a licensing regime delivered on laptops in rented convention halls owns the second one. The California February 2025 disruption was the field's first hard lesson in this; Washington is the second, and it lands on the inaugural day of the exam meant to be the fresh start. The questions it forces are structural: who certifies venue infrastructure, what redundancy is required, and who bears the risk when the room fails. Those are questions for supreme courts and bar associations, not just IT vendors. ANALYSIS: ours
What examinees are saying, and how to read week one honestly
The dominant examinee conversation in week one was about Washington, not the exam. CORROBORATED across platforms As that shifts, the early chatter deserves to be read with precision, because week-one exam talk is structurally unreliable in ways worth naming:
No one has a score. The NextGen reports on a 500–750 scale with passing lines set by each jurisdiction, and results will not exist for months. Every week-one claim about difficulty is a feeling calibrated against prep materials, not against performance. OFFICIAL scale; inference labeled
The first cohort has no baseline. There is no prior NextGen administration to compare against. "Harder than I expected" measures the taker's expectations (set by prep providers, ours included), not the exam against its own history. That makes first-cohort impressions uniquely interesting and uniquely uncalibrated at the same time.
Self-selection runs the room. The people most likely to post the night after a bar exam are the rattled and the triumphant. The quiet middle, most of the room, posts later or never. Any thread you read in week one is a sample of extremes.
Confidentiality mutes the specific. Examinees are bound to content confidentiality, so the most concrete reports are exactly the ones that can't and shouldn't be shared. What remains shareable (pacing, fatigue, logistics, interface behavior) is real signal, but it accumulates slowly and deserves patience.
So we'll do this the slow way, in public. We are collecting firsthand experience accounts (logistics, pacing, format mechanics, never content) directly: if you sat for this exam, support@butfor.ai reaches a human who reads every message. When independent accounts accumulate into genuine patterns, the synthesis publishes here and to the email list first, with every theme labeled by corroboration strength.
And here is what becomes knowable, when: the September make-up, if held, is the delivery chain's second test; it will show whether Washington's failure was a fluke of one venue or a fault line. Jurisdictions will release first-administration results on their own calendars, historically months after the exam. And the first pass-line decisions on the new 500–750 scale will tell the profession more about this exam than any week-one thread. We'll track each against primary sources as they land.
What this means if you're taking it in 2027
The short version: the format held, and not every room did. Preparation for this exam now includes preparing for its delivery, and the affected Washington takers face a genuinely hard three-way decision with different retention, financial, and career-timing profiles. We've written the full analysis as a separate deep report: the remedy calculus for affected takers, the venue and logistics questions worth asking your jurisdiction, and what week one changes (and doesn't) about how to prepare. It's free by email at butfor.ai.
But For is an independent publication of Ink Forms Rivers, d/b/a But For, and is not affiliated with, sponsored by, or endorsed by the National Conference of Bar Examiners or any bar examining authority. "NextGen UBE" is referenced solely to identify the examination. All information about the Washington administration is provided for context only; affected applicants should rely exclusively on official WSBA and NCBE communications for their options and deadlines.
But For · Be the cause.
If the exam described above is the one you're preparing for: But For's bench was built to it, format for format: 534 original questions, twelve integrated sets, all six performance tasks, AI-graded writing, and a ledger that trains your gaps. Founding access is $199, and the three-question calibration is free.